Pedro Domingos blasts OpenAI: looped transformers aren't more opaque, CoT transparency is fiction
pmddomingos · x · 2026-09-03
ML researcher Pedro Domingos publicly pushed back on OpenAI, arguing that looped transformers are no more opaque than regular ones and should simply be run to whatever depth is needed. He added that relying on chains of thought to guarantee reasoning transparency is "fiction to begin with." The post is a notable interpretability debate jab at OpenAI's architecture and narrative choices.
More from Models
- Leak: Astra won't be the best model of the year; a 'monster' is slated for end of year — ChrisGPT · 2026-09-03
- Startup Mostik bridges AI models via their weights, tops ARC-AGI 3 at 1/20 the cost — nordicinst · 2026-09-03
- Anthropic launches browser tool to detect Claude-made files via C2PA content credentials — btibor91 · 2026-09-03
- ByteDance's looped language models match 12B rivals at 1.4B size, with Bengio as co-author — peterjliu · 2026-09-03
- Anthropic Weakened Safety Filters, Signed an AI Cyberattack Warning Letter, Then Shipped Mythos 5.1 Anyway — AgentBlackVeil · 2026-09-03
- Marin 535B A23B Frontier-Scale Training Run Is Fully Livestreamed — Sentdex · 2026-09-03